This is the companion website to the paper by Sumeet Agarwal, Candida Vaz, Alok Bhattacharya and Ashwin Srinivasan (2010; with the same title as above). This page provides supplementary information to accompany the paper, along with an applet which allows the user to enter an RNA sequence for classification as a miRNA precursor or otherwise. The applet constructs the most likely secondary structure for the given sequence, using a context-sensitive hidden Markov model. Along with the structure, the CSHMM generates a likelihood score for the sequence being a pre-miRNA. The sequence is classified as positive or negative (i.e., as precursor or non-precursor) based on a threshold on this likelihood score. A stand-alone Java version of the CSHMM code, which can be used to scan contiguous genomic regions for pre-miRNA candidates, is also available: CSHMMCode_Basic.tar.gz (this requires WEKA, included in the tarball; please see readme.txt).
The applet requires Java. If you
don't have the Java Runtime Environment (IE users may not need it, as IE
generally comes packaged with the Microsoft JVM), you can get it here. The input to the
applet must be a RNA/DNA sequence consisting only of the four nucleotides (A,
C, G and U/T). Case doesn't matter. If the input sequence contains any
character other than the nucleotide letters, an error will be given. If the
input sequence is valid, the applet will display the constructed secondary
structure, the likelihood value, and the classification as a precursor or
non-precursor.
Note: As the input is supposed to be an RNA sequence, DNA sequences, if used as
input, will be treated as pre-transcription RNA sequences. This means that 'T'
will be treated just like 'U', and G-T pairing will be allowed, just as G-U
pairing is allowed for RNA sequences.
Please send any feedback to sumeet@iitd.ac.in.